The insulation state of cable can be split into well, bad, worse and fault. The estimation for cable state is a significant topic based on overhaul data, test data and monitor data of the cables. The decision tree is employed to classify the insulation state. The subtrees can be formed by all kinds of data, then the final decision tree is composed of the subtrees, by which the insulation state can be estimated. The application by SPSS with practical data of the cable is carried on. The simulation result for t(he insulation state estimation of cable shows the effectiveness of the approach.